Their approach uses file locking. Tasks have to be file-disjoint or agents overwrite each other. CAS uses git worktrees instead. Each agent gets its own full copy of the repo on its own branch. Three agents can edit the same file simultaneously and the supervisor merges everything back. No locks, no file-disjoint constraint.

The other problem: agents that say "done" when they're not. CAS has verification gates. Workers self-verify before closing (no TODOs, code is wired up, tests pass). Tasks enter pending_verification and workers can't claim new work until it clears. This was born out of pain. Without it, agents mark tasks done while leaving half-connected code everywhere.

GitHub: https://github.com/codingagentsystem/cas Website: https://cas.dev

How it works

`cas` launches a TUI with a supervisor. You give it an epic and it takes over: analyzes the tasks, figures out the right implementation phases based on dependencies, determines how many workers can run in parallel without file conflicts, and spawns them. Each task gets a demo statement describing the observable outcome ("User types query, results filter live"). This was the single biggest quality lever I found. Without it, agents default to building layers of plumbing that never connect to a working feature.

For inter-agent messaging, we reverse-engineered Claude Code's Team feature to build a push-based SQLite message queue. The previous version injected prompts by writing raw bytes into the terminal multiplexer. It worked, but barely. Now the supervisor sends a message and the worker's next MCP call picks it up. Clean and reliable.

Workers claim tasks via a lease system (renewed by heartbeat) to prevent double-claiming. The supervisor reviews completed work, merges the branch, syncs remaining workers, and assigns next tasks.

The TUI has side-by-side/tabbed views, session recording/playback, desktop notifications, and detach/reattach. Terminal emulation uses a custom VT parser based on Ghostty's.

Persistent context

CAS also runs as an MCP server (55+ tools) so agents remember things between sessions. 4-tier memory inspired by MemGPT. Rules that earn trust (Draft > Proven > Stale) and auto-sync to .claude/rules/. Full-text search via Tantivy BM25. Everything local in SQLite.

What's hard

Agent coordination is a distributed systems problem wearing a trenchcoat. Stale leases, zombie worktrees, agents that lie about completion. We've added heartbeats, verification gates, and lease expiry, but supervisor quality still varies with epic complexity. The honest answer is this is an ongoing arms race.

Getting started

curl -fsSL https://cas.dev/install.sh | sh cas init --yes && cas

~235K lines of Rust, 17 crates, MIT licensed. Happy to answer questions.

Comments URL: https://news.ycombinator.com/item?id=47341351

Points: 4

# Comments: 3

[1] https://cas.dev

I do understand anthropic's Tool Search helps with mcp bloat, but it's limited only to claude.

CMCP currently supports codex and claude but PRs are welcome to add more clients.

[1]https://blog.cloudflare.com/code-mode-mcp/ [2]https://github.com/assimelha/cmcp

With cmcp you get to use all thr available mcps, while still not bloating context.

The agent writes TypeScript to discover and call tools:

// search — find tools across all servers return tools.filter(t => t.name.includes("screenshot"));

// execute — call tools with full type safety await chrome_devtools.navigate_page({ url: "https://example.com" }); const shot = await chrome_devtools.take_screenshot({ format: "png" }); return shot;

Type declarations are auto-generated from each tool's JSON Schema, so the agent gets typed parameters for every tool. TypeScript is stripped via oxc and the JS runs in a sandboxed QuickJS engine (64 MB memory limit).

Adding servers works exactly like you'd expect — just prepend cmcp to any claude mcp add command from a README:

cmcp claude mcp add chrome-devtools npx chrome-devtools-mcp@latest cmcp install

Built in Rust with rmcp, rquickjs, and oxc. Inspired by Cloudflare's blog post on code-mode MCP.

What I found interesting building this: the biggest win isn't just fewer tokens — it's composability. An agent can chain calls across multiple servers in a single execution, which isn't possible with individual tool calls.

Comments URL: https://news.ycombinator.com/item?id=47099142

Points: 2

# Comments: 2

Comments URL: https://news.ycombinator.com/item?id=41520193

Points: 1

# Comments: 0

However, we've found them incredibly useful for other roles, especially when we're looking to fill positions like virtual assistants, social media managers, and similar jobs. So we built Activeinterview as a tool that we use internally + to sell as a SaaS.

This approach really shines for us in scenarios where we're sifting through a large volume of applications - think 100 or more. It's been a game-changer in helping us manage the initial stages of the hiring process more efficiently.

Would love to hear your thoughts or experiences on this!

Comments URL: https://news.ycombinator.com/item?id=40840780

Points: 2

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=37854058

Points: 2

# Comments: 0

I'm looking for one or two Senior React software engineers to join us at Nordbeam. We're a React/React Native consultancy based in Sweden but we work with remote developers all the time. If interested, apply through here: https://activeinterview.com/interviews/intrv_02uUKEPZcz5Xq1T...

Since fly deploys at the edge closer to the users, our response times for notifications are 20ms on average which is mind blowing since I haven’t spent more than 1 hour on infrastructure setup.

1: https://usenative.com

Comments URL: https://news.ycombinator.com/item?id=30011901

Points: 2

# Comments: 0

Please shares tips and experiences that might help

Comments URL: https://news.ycombinator.com/item?id=26034609

Points: 2

# Comments: 1